Separating subjective emotion from the perception of emotion-inducing stimuli: An fMRI study

www.elsevier.com/locate/ynimg

NeuroImage 33 (2006) 263 – 274

Separating subjective emotion from the perception of

emotion-inducing stimuli: An fMRI study

Amy S. Garrett a,* and Richard J. Maddockb

aStanford University School of Medicine, Department of Psychiatry, 401 Quarry Road, Stanford, CA 94305-5795, USAbUniversity of California, Davis Medical Center, Department of Psychiatry, CA 95814, USA

Received 15 April 2005; revised 3 May 2006; accepted 8 May 2006

fMRI was used to dissociate neural responses temporally associated

with the subjective experience of emotion from those associated with

the perception of emotion-inducing stimuli in order to better define the

emotion-related functions of the amygdala, lateral orbital frontal

cortex (OFC), and hippocampus. Subjects viewed aversive pictures

followed by an extended post-stimulus period of sustained subjective

emotion. Brain regions showing activation paralleling the period of

sustained subjective emotion were distinguished from those showing

activation limited to the period of aversive picture presentation.

Behavioral results showed that subjective ratings of emotion remained

elevated for 20 s after offset of the aversive pictures. fMRI results

showed that viewing aversive pictures activated the amygdala, lateral

OFC, and hippocampus. Subjective emotion (present both during and

after aversive pictures) was temporally associated with activation in the

right lateral OFC and left hippocampus but not the amygdala. Ratings

of subjective emotion were correlated with activation in the right

lateral OFC and left hippocampus. The results support direct

amygdala involvement in emotion perception but suggest that amyg-

dala activation is not temporally associated with subjective emotion

that occurs after the offset of emotion-related stimuli. The results are

consistent with a general role for the lateral OFC in monitoring or

reflecting on internal experience and show that hippocampal activation

is sustained during a period of subjective emotion, possibly related to

enhanced memory encoding for the aversive pictures.

D 2006 Elsevier Inc. All rights reserved.

Keywords: Emotion; fMRI; Amygdala; Orbitofrontal; Hippocampus;

Sustained

Introduction

The emotional response to an affectively relevant stimulus is

comprised of multiple processes, including perception, attention,

evaluation, memory, autonomic responses, cognitive responses,

behavior, and subjective experience (Cacioppo and Gardner,

1999). While medial temporal, prefrontal, and other brain regions

1053-8119/$ - see front matter D 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.neuroimage.2006.05.024

* Corresponding author.

E-mail address: [email protected] (A.S. Garrett).

Available online on ScienceDirect (www.sciencedirect.com).

have been consistently implicated in the emotional response,

important questions remain about the specific processes they

mediate.

The various components of an emotional response differ in their

temporal characteristics. Emotional processes involving attention

and perception are typically limited to the time the evoking

stimulus is present, while the subjective experience of emotion,

autonomic responses, cognitive responses, and behaviors such as

facial expressions are often sustained beyond the offset of the

evoking stimulus (Sirota et al., 1987; Bradley et al., 1996; Baker et

al., 1997; Pizzagalli et al., 1999; Garrett and Maddock, 2001).

These temporal characteristics might be used to identify brain

regions mediating these various components. Previously, we

showed that subjects report negative subjective emotion while

viewing aversive pictures and for about 16 s after the offset of

aversive pictures (Garrett and Maddock, 2001). Thus, while picture

perception is confined to the duration of picture presentation,

subjective emotion can be measured over a relatively sustained

period of time. Using this method, we can test whether regions

such as the amygdala, OFC, and hippocampus are involved

primarily in emotion perception and/or emotional responses that

parallel the timecourse of subjective emotional experience.

The amygdala appears to be critical for lending affective

significance to stimuli during perception (Adolphs and Tranel,

1999a,b; Adolphs and Tranel, 1999a,b; Adolphs et al., 1999;

Hamann et al., 1999; Anderson and Phelps, 2001; Amaral, 2002;

Amaral et al., 2003), but its role in the more sustained process of

subjective emotion is uncertain. Mood and subjective emotional

experience are generally unaffected in patients with amygdala

lesions (Anderson and Phelps, 2002), suggesting that these

functions are independent of the amygdala. In contrast, neuro-

imaging studies often show amygdala activation during induced

emotion (Phan et al., 2002; Levesque et al., 2003), and increased

amgydala activation has been shown in a group of 5 subjects

instructed to maintain a negative subjective emotion after picture

presentation (Schaefer et al., 2002). However, these studies have

not examined the unmanipulated subjective experience of emotion

as distinct from the perception of an emotion-inducing stimulus

throughout the entire brain.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274264

In addition to amygdala activation, hippocampal activation is

often reported in studies of emotion, primarily related to enhanced

encoding of emotional stimuli (McGaugh, 2000; Pare, 2003),

through the interaction between the amygdala and hippocampus

(Dolcos et al., 2004) (Richardson et al., 2004). Therefore, we might

predict that hippocampal activation would be parallel to amygdala

activation, that is, limited to the period of perception of emotion-

related stimuli.

A number of brain regions have been proposed to have a role in

the subjective experience of emotion, including the medial and

orbital prefrontal cortices, the insular cortex, and the somatosensory

cortex (Damasio, 1999; Mayberg et al., 1999; Lane, 2000; LeDoux,

2000; Craig, 2002; Hornak et al., 2003; Phillips et al., 2003).

Brodmann area (BA) 47, in the lateral OFC, is the prefrontal region

most commonly activated in studies of emotion (Steele and Lawrie,

2004). This region also has been implicated in emotional and non-

emotional meta-cognitive processes of self-monitoring and self-

reflection and emotion regulation (Hariri et al., 2000; Lieberman et

al., 2004). It has been linked to reflective awareness of subjective

emotion {Lieberman et al., 2004 #121} as well as self-reflection

associated with meta-memory phenomena (Maril et al., 2001; Kikyo

et al., 2002). Previous studies have suggested an association between

the lateral OFC and ratings of subjective responses in several

sensory modalities, including the subjective response to tastes

(Kringelbach et al., 2003; Small et al., 2003), odors (Anderson et al.,

2003), the affective component of pain (Rolls et al., 2003), and sad

feelings induced by watching films (Levesque et al., 2003). Taken

together, these findings suggest that the lateral OFC is activated

when subjects reflect on and assess a variety of subjective

experiences. Studies of emotion suggest that subjective emotion is

often accompanied by spontaneous self-reflection {Schooler, 2002

#129; Lieberman et al., 2004 #121}. If so, then the lateral OFC may

exhibit activation that is temporally associated with the subjective

experience of emotion, that is, observed during the presentation of

aversive pictures and also after the offset of the pictures.

The current fMRI study uses epochs of aversive pictures

followed by extended post-stimulus epochs of subjective emotion

to identify brain activation temporally associated with aversive

picture perception as well as those temporally associated with the

experience of subjective emotion. The hypothesis that amygdala

activation is associated with perception of emotion-related stimuli

but not subjective emotional experience predicts that its activation,

as well as that of the hippocampus, will be limited to the time the

aversive stimuli are present. The hypothesis that the lateral OFC

mediates spontaneous self-reflection that is temporally associated

with subjective emotion predicts that its activation will be observed

during the presentation of aversive pictures and also during the

subsequent epoch of continuing subjective emotion.

Fig. 1. Design of the emotion task. NES=n

Methods

Subjects

The Human Subjects Committee at University of California,

Davis, approved all protocols. Nine volunteers (4 male) were

recruited by advertisement from the local community and paid for

participation. All subjects gave informed consent and underwent a

screening interview to rule out psychiatric or neurological

conditions and medications affecting neural or cerebrovascular

function. All subjects reported that they were right-hand dominant.

Ages ranged from 27 to 50 (average = 38.67, standard deviation

(SD) = 8.7), education ranged from 12 to 18 years (average =

14.22, SD = 2.68) and ethnicity included 5 White, 2 Hispanic, 1

African American, and 1 African American/White individual.

Stimuli

Fifty-two pictures were selected from the International Affec-

tive Picture Set (Lang et al., 1997) and from additional highly

aversive pictures generously provided by our colleagues (Liberzon

et al., 2000) and previously described in detail (Garrett and

Maddock, 2001). Briefly, 26 aversive pictures were chosen to be as

unpleasant and arousing as possible, in order to evoke a lasting

emotional response, including scenes of violence, mutilation, and

injury. The remaining 26 pictures were chosen to be neutral, having

medium levels of arousal and valence. All pictures were chosen

based on published ratings and were equated on measures of

luminosity and complexity.

Emotion task

An abbreviated version of the Negative Emotional State (NES)

rating scale (Garrett and Maddock, 2001) was used by subjects to

indicate the intensity of their negative emotions while in the

scanner. Ratings for the abbreviated version ranged from 1 to 5,

where a rating of 1 corresponded to ‘‘no negative feelings’’ and the

scale descriptors ‘‘not upset, not uneasy, not bothered’’. A rating of

3 corresponded to ‘‘moderately negative feelings’’ and the

descriptors ‘‘uncomfortable, uneasy, bothered’’. A rating of 5

corresponded to ‘‘extremely negative feelings’’ and the descriptors

‘‘upset, shocked, queasy’’. Each subject practiced rating the

intensity of his/her current negative feelings following the viewing

of a sample set of neutral pictures.

For the emotion task, 20-s picture epochs were followed by 26-s

rating epochs (Fig. 1). During the picture epochs, subjects were

instructed to look carefully at the pictures and to let any emotions

arise naturally if they occurred. For the rating epochs, the subject

egative emotion scale; sec=seconds.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274 265

was instructed not to rate the pictures. Specifically, the subject was

told to rate his/her current negative feelings at that moment, whether

or not the feelings had been elicited by the previous pictures or his/

her own thoughts and memories. Subjective reports of the current

emotional state prompted by cues have been suggested to be a

relatively accurate estimate of ongoing experience, more accurate,

for example, than ratings referring to a previous time (Hurlburt and

Heavey, 2001). Finally, the subject was told that he/she would be

asked to make several ratings of emotion in succession, to assess

whether or not their emotions changed over time. The subject was

not biased to report changes in ratings over time, as we explained

that ‘‘sometimes emotions do not change over time, and sometimes

emotions do change over time. Please let us know what you

experience’’.

The task was presented, and ratings were collected using

Eprime software (Psychology Software Tools, Pittsburgh, PA).

During each picture epoch, 4 pictures were presented at the rate of

5 s per picture. During the rating epochs, 4 consecutive ratings

were collected, at the rate of 1 rating per 6 s. During the rating

epochs, a fixation cross was presented. The fixation cross

intermittently turned to a star (*) to signal the subject to make a

rating by pressing a button. In addition, 2 s was allowed at the

beginning of each rating period in order to give the subject a brief

interval to transition to the rating epoch. Six aversive and six

neutral picture epochs were presented. Each picture epoch was

followed by a rating epoch. An additional neutral picture epoch

and rating epoch were used as an initial baseline but were not

analyzed, as they were designed to allow the subject to acclimate to

the scanner and the task.

Imaging protocol

Images were obtained with a 1.5-T magnetic resonance imaging

system (Signa Advantage: General Electric, OS version 8.4) with a

GE birdcage head coil. For anatomical localization in each subject,

a high-resolution Fast Spin Echo scan was acquired in the coronal

plane (TR = 3100; effective TE = 17 and 36; matrix = 256 � 256;

FOV = 22 cm; slice thickness = 6 mm with a 2 mm gap; 24 slices).

Functional images were acquired using a T2*-weighted gradient

recalled echo planar imaging (EPI) pulse sequence with shim (TR =

2000 ms; effective TE = 40 ms; flip angle = 90-; matrix = 64 � 64;

FOV = 22 cm; slice thickness = 6 mm with 2 mm gap; 18 coronal

slices). Activation was detected by the blood oxygenation level

dependent (BOLD) contrast. Head motion was constrained by foam

padding.

Picture ratings

Following the end of the scanning session, subjects rated each

picture using the Self-Assessment Mannikin (SAM, Bradley and

Lang, 1994). Each picture was rated on dimensions of valence and

arousal in order to verify that the subjects perceived the pictures as

intended.

Functional image analysis

Images were reconstructed using a Fourier transform-based

algorithm with removal of N/2 ghost artifacts (Buonocore and Gao,

1997). SPM99 (http://www.fil.ion.bpmf.ac.uk/spm) was used for

image analysis, including adjustment for timing of slice acquisi-

tion. Realignment for movement correction used a least squares

minimization without higher order corrections for spin history.

Images were normalized to the Montreal Neurological Institute

template and resampled by sinc interpolation into 2 � 2 � 2 mm

voxels. Spatial smoothing was performed using a 4-mm Gaussian

kernel.

Whole-brain statistical analyses were performed using the

theory of Gaussian random fields (Holmes and Friston, 1998).

For each subject, confounding effects of fluctuation in global mean

were removed by proportional scaling, where each voxel was

scaled by the global mean at each time point. Low-frequency noise

was removed using a high-pass filter. A canonical hemodynamic

response function was used to model the BOLD response.

Comparisons of interest were performed for each subject by

computing voxel-wise t statistics normalized to Z scores to provide

a statistical measure of activation independent of sample size. A

random effects model was used to combine individual subject data

into a group analysis. This model estimates the error variance for

each condition of interest across subjects, rather than across scans,

providing better generalization to the population. Statistical

significance was determined as clusters that passed a joint

threshold of height (Z > 1.67; P < 0.05) and extent (cluster size

corresponding to P < 0.05) after correction for multiple compar-

isons (Poline et al., 1997).

Analysis of the emotion task

Brain regions associated with the perception of negative

emotional stimuli were identified as those activated during the

aversive picture epochs (compared to the neutral picture epochs)

but not during the post-aversive rating epochs (compared to the

post-neutral rating epochs). Regional activations temporally

associated with the subjective experience of negative emotion

were identified as those activated in both the aversive picture

contrast and the post-aversive rating contrast. This conjunction

analysis was used instead of a reference vector that modeled the

emotional response across the entire pictures and ratings epochs in

order to avoid biasing the results toward regions activated during

the picture epochs. Strong activation during the picture epochs

leads to false positive activation when considering the picture and

ratings epochs together. Therefore, the conjunction analysis is the

most conservative method for identifying activation temporally

associated with subjective emotion both during and after the

picture blocks.

This analysis is different than the mixed blocked/event-related

design described previously to separate transient and sustained

activity in fMRI (Visscher et al., 2003). First, responses to the

pictures were not analyzed as events because the goal was to build

a negative emotional response across the block and not to address

the question of responses to individual aversive pictures. Secondly,

the established method of analyzing sustained activation is

designed to isolate components of the hemodynamic response that

last throughout the series of pictures but are not maintained beyond

the picture block. The current study, by contrast, is interested in

identifying activation that continues in the absence of picture

stimuli, just as the subjective experience of negative emotion

continues beyond the termination of picture stimuli.

To increase our sensitivity to activation associated with

negative subjective emotion, the rating block having the lowest

NES ratings for each individual subject was excluded from the

analysis (Canli et al., 2002). Also, we truncated emotion rating

blocks to 20 s to isolate the time of elevated negative emotion.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274266

Therefore, the first 20 s of the aversive emotion rating epochs were

compared to the first 20 s of the neutral emotion rating epochs.

Twenty seconds was chosen based on the ratings of subjective

emotion reported by the subjects in this study (reported below).

In a separate analysis, correlations between ratings and brain

activation were performed in SPM99 by specifying each subject’s

vector of NES ratings as a regressor in the linear model. Activation

related to the ratings was chosen as the contrast of interest.

Results

Picture ratings

Repeated measures ANOVA confirmed that subjects rated the

aversive pictures as significantly higher in arousal (F(1,46) =

158.28, P < 0.0001) and significantly more negative in valence

(F(1,46) = 102.14, P < 0.0001) compared to the neutral pictures.

Subjective emotion ratings

Fig. 2A shows the average timecourse of the NES ratings over

the course of the experiment for all subjects combined. The

average NES rating following the aversive pictures was 3.39 (SD =

1.02) and following neutral pictures was 1.41 (SD = 0.75),

showing that NES ratings during the aversive rating periods were

significantly greater than NES ratings during the neutral rating

periods (Wilcoxon Signed Rank Z = �11.5, P < 0.0001). There

was no habituation of NES ratings over the course of the

experiment, that is, the average rating per aversive block did not

change significantly (repeated ANOVA, F(8,5) = 0.41, P < 0.84).

NES ratings remained elevated for 20 s following presentation of

the aversive pictures, confirming that subjects experienced a

sustained subjective emotional response following the aversive

pictures but not following the neutral pictures. Compared to the

first rating of the rating epoch, the second NES ratings declined an

average of 4% (12 s into the epoch), the third NES rating declined

33% (18 s into the epoch), and the fourth rating declined 52% (24 s

into the epoch).

Fig. 2. Group mean ratings of negative subjective emotion collected follow

Brain activation during the emotion task

Brain regions significantly activated during the presentation of

aversive pictures compared to neutral pictures are illustrated in Fig. 3

and listed in Table 1. Regions include the bilateral amygdalae,

hippocampi, lateral OFC, and visual cortices. Brain regions

significantly activated during the emotion rating blocks following

aversive compared to neutral pictures are illustrated in Fig. 3 and

listed in Table 2. Regions include the hippocampus, lateral OFC,

thalamus, and cerebellum.

Brain regions that were significantly activated during both the

aversive picture block and the subsequent aversive rating blocks

(temporally associatedwith subjective emotion) are shown in Fig. 3 as

the overlap of activation from each contrast and are listed in Table 2.

These include the left hippocampus, right lateral OFC, thalamus, and

cerebellum. Brain regions activated during the aversive picture block

but not the subsequent rating block (temporally associated with

emotional perception) included the amygdala and visual cortex.

Brain regions having activation that increases with increasing

negative emotion ratings (following both aversive and neutral

pictures) are listed in Table 3. Regions include the right lateral

OFC, hippocampus, anterior cingulate cortex, thalamus, and

cerebellum (Fig. 4).

Timecourse graphs

As a post hoc qualitative verification of the results found in the

whole-brain analysis, the raw timecourse of voxels in the amygdala,

hippocampus, OFC, and visual cortex were plotted. Because of

vasculature differences between brain regions, each region is

compared to itself in order to evaluate the relative timecourse of

response as prominent during picture perception or paralleling

subjective emotional experience. To obtain the plots of OFC and

hippocampus, a sphere was centered on the maximally activated

voxel in the overlap of activation during picture and emotion rating

epochs. For the visual cortex and amygdala, which did not show

activation during the emotion rating epochs, the sphere was chosen

to capture the voxels of maximal activation during the picture

epochs. The location of the center of the sphere for each region are

ing the neutral pictures (N) and following the aversive pictures (A).

Fig. 3. Activation associated with viewing aversive pictures (compared to neutral pictures) is shown in red. Activation associated with rating subjective negative

emotion following the aversive pictures (compared to rating emotion following the neutral pictures) is shown in blue. Sustained activation that is present in both

the aversive picture and the subsequent aversive rating epochs are areas of overlap of red and blue, circled here in green. All clusters are significant at p<0.05

corrected. Color scale represents T values. Apics = aversive pictures; Npics = neutral pictures; Arate = aversive ratings; Nrate = neutral ratings.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274 267

as follows: visual cortex: (�42,�52,�18) amygdala: (�22, 0,�20)hippocampus: (�24, �22, �15) OFC: (46, 36, �16). A 4-mm

diameter sphere was centered on that voxel on each subject’s

individual contrast map, and the average timecourse from all voxels

in the sphere was extracted. A sphere was used in order to isolate

each brain region from within the larger cluster of activation. The

Table 1

Areas of significant activation to aversive pictures compared to neutral pictures

Region (Brodmann’s area) # Voxels (corrected

P of cluster)

Z score of peak

voxel in region

Talairach coordinates,

X, Y, Z

Uncorrected P of

peak voxel in region

L middle occipital gyrus (19) 3531 4.41 �48, �77, 15 0.001

L inferior occipital gyrus (19) (P = 0.001) 4.19 �36, �72, �3 0.001

L fusiform gyrus (19) 3.47 �40, �66, �8 0.001

L inferior temporal gyrus (37) 3.49 �50, �62, �2 0.001

L middle temporal gyrus (37/21) 3.62 �48, �54, 6 0.001

R inferior temporal gyrus (37) 4366 4.41 51, �64, 3 0.001

R inferior occipital gyrus (19) (P = 0.001) 3.92 42, �74, �8 0.001

R middle occipital gyrus (18/19) 4.36 30, �78, 24 0.001

R fusiform gyrus (19) 4.03 46, �67, �10 0.001

R cerebellum 2.78 40, �44, �26 0.003

R middle temporal gyrus (37) 3.92 50, �56, 1 0.001

R superior temporal sulcus (21/22) 2.23 55, �48, 19 0.013

L amygdala 4424 4.21 �22, 3, �17 0.001

R amygdala (P = 0.001) 2.87 20, �2, �10 0.002

L parahippocampal gyrus 2.95 �14, �31, �5 0.002

L hippocampus 2.44 �30, �16, �16 0.007

R hippocampus 3.11 22, �10, �15 0.001

Midbrain 3.63 6, �32, �12 0.001

R thalamus 3.30 4, �15, 4 0.001

R/L hypothalamus 3.98 2, �2, �5 0.001

R putamen 2.64 26, 13, �4 0.004

L putamen 3.72 �24, 11, �6 0.001

R caudate 2.83 12, 12, 1 0.002

L caudate 4.10 �16, 17, �1 0.001

R anterior insula 2.86 38, 11, �4 0.002

R lateral OFC (11/47) 3.75 40, 29, �10 0.001

L lateral OFC (11/47) 3.84 �34, 25, �8 0.001

R OFC (10/11) 3.71 26, 26, �15 0.001

L OFC (10/11) 3.39 �38, 34, �15 0.001

OFC = orbital frontal cortex; inf = inferior; R = right; L = left.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274268

timecourses were averaged across blocks, i.e., the aversive blocks

(including both picture and subsequent ratings blocks) were

averaged into a single timecourse per subject, as were the neutral

blocks. Individual subject timecourse data were normalized, to

remove differences in relative intensity values, by dividing each

point in the timecourse by the average of the first 2 timepoints (4 s) of

that same timecourse, which corresponds to a baseline. This method

is commonly used in the analysis of scalp-recorded event-related

potentials (Dawson, 1954). After normalization, all subjects’

average timecourse data were combined into a group average and

graphed as the change in intensity over time.

Table 2

Areas of significant activation during negative emotion ratings following aversive compared to neutral pictures

Region (Brodmann’s area) # Voxels (corrected

P of cluster)

Z score of peak

voxel in region

Talairach coordinates

X, Y, Z

Uncorrected P of

peak voxel in region

Cerebellar vermis 1786 4.35 4, �70, �32 0.001

R/L cerebelluma (P = 0.001) 3.94 �6, �78, �10 0.001

R/L lingual gyrus (18) 3.14 4, �82, �4 0.001

R/L thalamusa 624 3.82 4, �29, 1 0.001

L hippocampusa (P = 0.03) 2.77 �14, �22, �7 0.003

R/L hypothalamusa 2.90 �2, �4, �7 0.002

R lateral OFC (11/47)a 841 3.40 50, 36, �14 0.001

R OFC (11) (P = 0.004) 2.94 30, 48, �6 0.002

R inferior frontal gyrus (45/47) 2.89 48, 36, �5 0.002

R frontal operculum/anterior insula 2.54 48, 21, 1 0.006

OFC = orbital frontal cortex; R = right; L = left.a Region showing sustained activation (activation overlapped that seen in the aversive compared to neutral pictures contrast).

Fig. 5 shows the graph of the timecourses. For all regions,

activation is delayed about 3 TR, consistent with a standard

hemodynamic lag. Note that activation in the visual cortex is

transient for both the aversive and neutral blocks. Amygdala

activation is elevated for a longer time during the aversive

compared to neutral blocks but still declines approximately at the

end of picture presentation during the aversive blocks. However,

hippocampal activation is sustained well into the rating epochs for

the aversive but not neutral blocks. Activation in the lateral OFC is

also sustained into the rating epoch for the aversive compared to

the neutral blocks. These graphs confirm the results of the

Table 3

Areas of activation significantly correlated with ratings of negative emotions

Region (Brodmann’s area) # Voxels (corrected P of cluster) Z score of peak voxel Talairach coordinates X, Y, Z P value of peak voxel

Cerebellar vermis 3336 4.41 4, �71, �30 0.001

R cerebellum (P = 0.001) 4.00 32, �42, �26 0.001

R hippocampus 4.02 16, �20, �12 0.001

L hippocampus 3.33 �14, �16, �9 0.001

R thalamus 3.09 8, �19, 8 0.001

L thalamus 3.05 �6, �11, 8 0.001

R lateral OFC (11/47) 810 4.17 26, 52, �18 0.001

R anterior insula (P = 0.01) 3.29 50, 32, �17 0.001

L precentral gyrus (4) 1277 3.96 �36, �7, 56 0.001

L postcentral gyrus (1/2) (P = 0.001) 3.39 �32, �13, 47 0.001

L middle frontal (6) 3.16 �28, 9, 59 0.001

L medial frontal (6/32) 756 3.57 �8, 12, 49 0.001

R medial frontal (6/32) (P = 0.015) 4.58 6, 12, 40 0.001

R anterior cingulate (24) 3.02 2, 23, 32 0.001

L anterior cingulate (24) 1.80 �8, 9, 33 0.036

OFC = orbital frontal cortex; R = right; L = left.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274 269

quantitative analysis and illustrate that the lateral OFC and

hippocampal activations begin during the presentation of the

aversive pictures, and remain above baseline concurrent with the

subjective experience of negative emotion.

Discussion

This study shows that the neural response to highly arousing,

unpleasant (aversive) pictures can be separated into activation

temporally associated with emotional perception and activation

temporally associated with the subjective experience of emotion.

We identified brain regions in which the response to aversive

pictures was limited to the period of stimulus presentation and

brain regions in which the response was sustained beyond the

period of stimulus presentation while subjects continued to

experience negative emotion. During the presentation of aversive

pictures, activation was seen in predicted regions, including the

visual cortex, amygdala, hippocampus, and lateral OFC. During

the subsequent epoch, while subjects’ ratings indicated an ongoing

negative emotional state, activation continued in the lateral OFC

and the hippocampus but not in the amygdala. These data suggest

that the amygdala is primarily activated during the perception of

emotion-related stimuli, but not during subjective emotion that

occurs after the offset of the emotion-related stimuli. Activation in

the lateral OFC during the aversive pictures continued after the

offset of the pictures while subjects reported ongoing subjective

emotion, indicating that this region may have a role in subjective

emotion. However, other studies suggest that this region is more

likely involved in the self-monitoring or self-reflection that often

accompanies subjective emotion. Finally, the hippocampus showed

sustained activation throughout the presentation of stimuli and the

subjective emotion rating epochs and may be related to enhanced

encoding of aversive stimuli.

The findings suggest that the amygdala plays a relatively

transient role in the emotional response to aversive pictures.

Amygdala activation increased during the perception of aversive

pictures but did not persist during subjective reports of sustained

negative emotion. Furthermore, ratings of subjective emotion

collected after picture offset did not show a correlation with

amygdala activation. While the role of the amygdala is emotional

processes in general is well established, the role of the amygdala

specifically in subjective emotion has been debated. Previous

neuroimaging studies have shown amygdala activation during a

variety of emotion-related paradigms (see Phan et al., 2002 for a

review), including during induced emotional states (Levesque et

al., 2003), correlated with ratings of subjective emotion, including

pain (Ketter et al., 1996; Schneider et al., 2001), and amygdala

activation that was up or down-modulated while subjects con-

sciously up or down regulated subjective negative emotion

(Ochsner et al., 2004). However, these studies have not separated

the effects of emotion-inducing stimuli from those of subjective

emotion. Schaefer et al. (2002) observed increased amygdala

activation while subjects consciously maintained a negative mood,

in the absence of negative picture stimuli. However, this study did

not address the role of the amygdala during the natural timecourse

of subjective emotion. Moreover, human subjects with lesions of

the amygdala report normal subjective experience of emotion

(Anderson and Phelps, 2002) and amygdala lesions in humans

impair the perception of fearful facial expressions but do not

disrupt the expression of fear (Anderson and Phelps, 2000). The

current study supports the hypothesis that the amygdala partic-

ipates in perception of emotion-related stimuli but is not directly

involved in the subjective negative emotion that persists after the

offset of the evoking stimulus.

As predicted, the lateral OFC exhibited activation during

presentation of the aversive pictures that continued during the

post-stimulus period, during which subjects reported ongoing

negative emotion. Further, individual ratings of emotion were

significantly correlated with activation in this region. Together,

these findings suggest that the lateral OFC mediates some

component(s) of the emotional response that is temporally

associated with the subjective experience of emotion. Neuroana-

tomical evidence supports this idea, as the lateral OFC receives

sensory information, insular projections, and limbic input from the

amygdala and entorhinal cortex (Price, 1999) and therefore could

integrate, monitor, or regulate these components as part of a

multifaceted emotional response.

Many neuroimaging studies have shown activation of the

lateral OFC in response to emotion-inducing stimuli. In a recent

meta-analysis, Steele and Lawrie (2004) reviewed 30 published

experiments showing emotion-related activation along the orbital,

Fig. 4. Activation correlated with ratings of subjective negative emotion during the rating epochs (following both neutral and aversive pictures). All clusters are

significant at p < 0.05, corrected. Color scale represents T values.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274270

lateral, and dorsal surfaces of the prefrontal cortex. Among these

regions, BA 47 (in the lateral OFC) was the area most consistently

activated during induced emotion. Phillips et al. (2003) proposed

that this region contributes to the subjective experience of

emotion, and other investigators have suggested that its role

includes regulating the subjective experience of emotion (Eisen-

berger, 2003), perhaps by down-regulating activation of the

amygdala (Ochsner et al., 2004).

However, the lateral OFC may have a more general role in

making judgments that require reflecting on subjective feelings,

Fig. 5. Timecourses of activation in the left hippocampus, right lateral OFC, amygdala, and visual cortex during aversive, compared to neutral, pictures and

ratings. This graph is not adjusted for the hemodynamic delay, thus the hemodynamic response begins approximately 3 TR after the neural response. (a)

Activation that is transient relative to the comparison condition includes the visual cortex and left amygdala. (b) Activation that is sustained relative to the

comparison condition includes the left hippocampus and right lateral OFC.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274 271

including value judgments {Zysset, 2002 #85}, feelings of sadness

(Levesque et al., 2003), enjoyment while eating chocolate (Small et

al., 2001), pleasant and painful touch (Rolls et al., 2003), and taste

(Kringelbach et al., 2003). Also, neuroimaging studies without an

affective component show activation in BA 47 related to meta-

memory phenomena (Maril et al., 2001; Kikyo et al., 2002), active

retrieval in working memory (Wager and Smith, 2003), long-term

memory (Ranganath et al., 2003; Wig et al., 2004), semantic

retrieval (Wagner et al., 2001), and reasoning tasks (Goel et al.,

2004). Together, these studies suggest that BA 47 mediates a range

of executive functions, including reflecting on current subjective

experience, and that these functions can be flexibly recruited to

serve either emotional or non-emotional processes.

The subjective experience of emotion has been conceptualized

as including both phenomenal experience (specific feelings and

sensations) and self-reflection (meta-cognitive appraisal of phe-

nomenal experience) (Lane et al., 1997; Schooler, 2002; Lieber-

man et al., 2004). Although self-reflection does not accompany

all emotional responses, it is often spontaneously evoked during

strong, negative emotional responses and may have an important

role in the regulation of negative emotional states (Schooler,

2002; Lieberman et al., 2004). Lieberman et al. (2004) proposed

that the self-reflection evoked by negative emotion is mediated by

the right ventrolateral prefrontal cortex. The current results are

consistent with this formulation. It is worth noting that our design

allowed for both spontaneous self-reflection (accompanying

subjective emotion) and deliberate self-reflection (while making

self-ratings). However, the latter occurred only during the rating

epoch. Therefore, only self-reflection that spontaneously accom-

panied subjective emotion would have been captured by our

conjunction analysis. Experiments specifically designed to exam-

ine the role of self-reflection in subjective emotion will be

necessary to test the hypothesis that the right lateral OFC

mediates this process.

A follow-up analysis of data from our laboratory confirms that

the lateral OFC participates in non-emotional processes as well as

emotion-related evaluation. To test this, a second experiment,

called the memory-rating task, used the same subjects and a

similar experimental design. It was implemented as a control

experiment to identify activation related to the meta-cognitive

process of deliberately reflecting on and rating a non-emotional

internal experience and to determine whether activation during the

rating epochs was related to memory retrieval of the previously

presented pictures. The timing of the memory-rating task was

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274272

identical to the emotion task. Subjects viewed blocks of neutral

pictures, followed by blocks in which they made ratings. For the

picture epochs, subjects were instructed to look carefully at the

pictures but not to make any specific effort to remember them, as

they would not be tested on their memory of the pictures. For the

rating epochs, subjects were instructed to rate the clarity of their

memory at that moment for any of the previously shown pictures,

without explicitly trying to remember the pictures. These

instructions were used to discourage effortful strategies of

encoding and retrieval and to direct the subject to rate the clarity

of memories retrieved with minimal effort. The subjects were told

that they would be asked to make several ratings in succession, so

that we would know whether or not the clarity of their memory

changed over time. Picture/Rating epochs were compared with

epochs in which the subject was shown unrecognizable scrambled

pictures and then rated his/her subjective emotion using the NES

scale. Results showed that the lateral OFC (but not the amygdala

or hippocampus) was activated while subjects performed the non-

emotional memory-rating task. These results support our hypoth-

esis that BA 47 activation is not specific to emotion-related

processes but is related to the process of self-reflection, including

the deliberate self-reflection that occurs during the rating task and

the spontaneous self-reflection that occurs during subjective

emotion. The hippocampus was not activated during the rating

epochs, possibly because subjects were not asked to try to

remember the pictures and therefore no effortful encoding took

place, and the pictures were not emotion-related or otherwise

highly interesting.

We did not observe sustained medial prefrontal or anterior

cingulate cortex (ACC) responses to the aversive pictures.

However, ratings of subjective emotion were significantly corre-

lated with activation of the ACC. Several studies have suggested

that the ACC has an important role in the subjective experience of

negative emotion (Mayberg et al., 1999; Lane, 2000; Hornak et al.,

2003). One study (Taylor et al., 2003) showed that the process of

rating emotional pictures increased medial prefrontal and ACC

activation compared to passive viewing. Another (Lane et al.,

1998) reported that attention to subjective experience is associated

with increased anterior cingulate activation. The lack of sustained

ACC activation in this study may indicate variability in its

response to different experimental conditions or insufficient

statistical power. However, note that we cannot comment on the

lack of activation in portions of the ventral prefrontal cortex. An

analysis of signal dropout due to susceptibility artifact showed that

signal loss in the anterior ventromedial PFC extends to approx-

imately Z = �15 in the anterior PFC and about Z = �10 in the

posterior PFC. This is in line with the signal loss observed in other

studies. However, it would interfere with detecting activation in

the ventral portion of BA 25 (the subgenual and subcallosal

cortices), as well as more anterior parts of the ventromedial

prefrontal cortex.

Our finding of hippocampal activation is consistent with earlier

studies of negatively valenced emotional stimuli (Phillips et al.,

1998; Lane, 2000; Maratos et al., 2001; Williams et al., 2001;

Maddock et al., 2003; Phillips et al., 2003). However, this is the

first demonstration of emotional stimulus-induced hippocampal

activation that is sustained following the offset of the inducing

stimulus. Activation of the hippocampal region has been associ-

ated with a subsequent memory effect (better subsequent memory

for stimuli evoking greater activation at encoding) (Brewer et al.,

1998; Fernandez et al., 2002; Strange et al., 2002; Davachi et al.,

2003; Stark and Okado, 2003; Jackson and Schacter, 2004).

Animal, lesion, pharmacological and imaging studies show that

modulation of hippocampal activity by the amygdala mediates

enhanced memory for emotionally arousing stimuli (Anderson and

Phelps, 2001; Hamann, 2001; Kilpatrick and Cahill, 2003).

Although we did not assess memory in this study, the hippocampal

and amygdala activation may reflect enhanced memory encoding

for the aversive pictures. If so, our findings suggest that the

emotional memory processes mediated by the amygdala and

hippocampus may have different temporal dynamics, with only

hippocampal activity sustained during the post-stimulus period.

Fear conditioning studies suggest contrasting functions for the

amygdala and the hippocampus, with associations between

specific stimuli mediated by the amygdala and contextual

associations mediated by the hippocampus (Maren, 2001;

Matus-Amat et al., 2004). Explicit memory for emotional stimuli

and events may involve similar interactions between the

amygdala and hippocampus (McGaugh et al., 1996; Cahill and

McGaugh, 1998; Richardson et al., 2004). Contextual information

typically persists after the offset of aversive stimuli, as was the

case in our experiment. Encoding of contextual information may

have contributed to the sustained activation observed in the

hippocampus.

In the current study, the cerebellum was found to be activated

during picture perception epochs as well as emotion rating epochs

and to be correlated with ratings of negative emotion. This is

consistent with the suggestion that the cerebellum is involved in

some way in the subjective experience of emotion. Previous studies

of emotion have observed cerebellar activation, for example, in

association with happy moods in men (Habel et al., 2005), in

mothers viewing photos of their infants (Nitschke et al., 2004), and

in bereaved women viewing photos of the deceased (Gundel et al.,

2003). In addition, cerebellar lesions in humans can produce

emotional flattening and impaired autonomic activity to negative

stimuli (Annoni et al., 2003). However, no studies have sought to

clarify the specific role of the cerebellum in normal human emotion.

A recent paper by Sacchetti et al. (2005) reviews evidence that the

cerebellum participates in fear learning, and that inactivation of the

cerebellar vermis in animals during the consolidation of fear

conditioning impairs fear memory, through its interconnections

with the hypothalamus, amygdala, and hippocampus. These studies

suggest that cerebellar activation in the current study may be linked

to the subjective experience of emotion as part of the autonomic

response to negative emotional states.

Several limitations of this study should be noted. The sample

size is small and includes variations in age, gender, race, and

education level that could decrease our chances of detecting

significant activation. In particular, gender differences in

responses to emotion-related stimuli have been observed (Wager

et al., 2003). However, the small sample size did not prohibit

testing our hypotheses, as significant activation was found in all

contrasts. It is not clear why activation was seen in the left

hemisphere amygdala and hippocampus and in the right

hemisphere lateral OFC. A review by Wager et al. (2003)

concludes that there is no consistent pattern of lateralization of

activation to emotion-related stimuli. Combined with the small

sample size of the current study, we hesitate to interpret the

hemispheric patterns seen in the current study.

Although the timing of activation in the right lateral OFC

paralleled that of subjective emotion and the magnitude of

activation correlated with ratings of subjective emotion, we cannot

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274 273

be certain that this activation indicates a single neural process

temporally associated with the subjective experience of emotion. It

is possible that right lateral OFC activation during the emotion

rating period indicated only the explicit self-rating task. However,

activation in the same location was seen while subjects viewed the

emotional pictures without any explicit self-rating and has been

widely reported in other studies of emotion-inducing stimuli

without explicit self-rating tasks (Steele and Lawrie, 2004).

Requiring subjects to rate their subjective experience during

scanning may have diminished some components of their

emotional response (Liberzon et al., 2000; Lieberman et al.,

2004). Adding physiological measures of emotion to subjective

reports would provide a more complete assessment of emotional

responses. Furthermore, subjective ratings of emotion are an

imprecise measure of the emotional state of the subject due to

variability in reporting skills. However, cued reports of current

emotional state are more reliable and accurate than retrospective or

summary ratings (Hurlburt and Heavey, 2001), and subjective

ratings are essential for experimental investigations of the

subjective experience of emotion.

References

Adolphs, R., Tranel, D., 1999. Intact recognition of emotional prosody

following amygdala damage. Neuropsychologia 37 (11), 1285–1292.

Adolphs, R., Tranel, D., 1999. Preferences for visual stimuli following

amygdala damage. J. Cogn. Neurosci. 11 (6), 610–616.

Adolphs, R., Tranel, D., et al., 1999. Recognition of facial emotion in nine

individuals with bilateral amygdala damage. Neuropsychologia 37 (10),

1111–1117.

Amaral, D.G., 2002. The primate amygdala and the neurobiology of social

behavior: implications for understanding social anxiety. Biol. Psychiatry

51 (1), 11–17.

Amaral, D.G., Behniea, H., et al., 2003. Topographic organization of

projections from the amygdala to the visual cortex in the macaque

monkey. Neuroscience 118 (4), 1099–1120.

Anderson, A.K., Phelps, E.A., 2000. Expression without recognition:

contributions of the human amygdala to emotional communication.

Psychol. Sci. 11 (2), 106–111.

Anderson, A.K., Phelps, E.A., 2001. Lesions of the human amygdala impair

enhanced perception of emotionally salient events. Nature 411 (6835),

305–309.

Anderson, A.K., Phelps, E.A., 2002. Is the human amygdala critical for the

subjective experience of emotion? Evidence of intact dispositional

affect in patients with amygdala lesions. J. Cogn. Neurosci. 14 (5),

709–720.

Anderson, A.K., Christoff, K., et al., 2003. Dissociated neural representa-

tions of intensity and valence in human olfaction. Nat. Neurosci. 6 (2),

196–202.

Annoni, J.M., Ptak, R., et al., 2003. Decoupling of autonomic and

cognitive emotional reactions after cerebellar stroke. Ann. Neurol. 53

(5), 654–658.

Baker, S.C., Frith, C.D., et al., 1997. The interaction between mood and

cognitive function studied with PET. Psychol. Med. 27 (3), 565–578.

Bradley, M.M., Lang, P.J., 1994. Measuring emotion: the self-assessment

manikin and the semantic differential. J. Behav. Ther. Exp. Psychiatry

25 (1), 49–59.

Bradley, M.M., Cuthbert, B.N., et al., 1996. Picture media and emotion:

effects of a sustained affective context. Psychophysiology 33 (6),

662–670.

Brewer, J.B., Zhao, Z., et al., 1998. Making memories: brain activity that

predicts how well visual experience will be remembered. Science 281

(5380), 1185–1187.

Buonocore, M.H., Gao, L., 1997. Ghost artifact reduction for echo planar

imaging using image phase correction. Magn. Reson. Med. 38 (1),

89–100.

Cacioppo, J.T., Gardner, W.L., 1999. Emotion. Annu. Rev. Psychol. 50,

191–214.

Cahill, L., McGaugh, J.L., 1998. Mechanisms of emotional arousal and

lasting declarative memory. Trends Neurosci. 21 (7), 294–299.

Canli, T., Sivers, H., et al., 2002. Amygdala response to happy faces as a

function of extraversion. Science 296 (5576), 2191.

Craig, A.D., 2002. How do you feel? Interoception: the sense of the phy-

siological condition of the body. Nat. Rev., Neurosci. 3 (8), 655–666.

Damasio, A.R., 1999. The Feeling of What Happens: Body and Emotion in

the Making of Consciousness. Harcourt Brace, New York.

Davachi, L., Mitchell, J.P., et al., 2003. Multiple routes to memory: distinct

medial temporal lobe processes build item and source memories. Proc.

Natl. Acad. Sci. U. S. A. 100 (4), 2157–2162.

Dawson, G.D., 1954. A summation technique for the detection of small

evoked potentials. Electroencephalogr. Clin. Neurophysiol., Suppl. 6

(1), 65–84.

Dolcos, F., LaBar, K.S., Cabeza, R., 2004 Jun 10. Interaction between the

amygdala and the medial temporal lobe memory system predicts better

memory for emotional events. Neuron 42 (5), 855–863.

Fernandez, G., Klaver, P., et al., 2002. Human declarative memory

formation: segregating rhinal and hippocampal contributions. Hippo-

campus 12 (4), 514–519.

Garrett, A.S., Maddock, R.J., 2001. Time course of the subjective emotional

response to aversive pictures: relevance to fMRI studies. Psychiatry

Res. 108 (1), 39–48.

Goel, V., Makale, M., et al., 2004. The hippocampal system mediates

logical reasoning about familiar spatial environments. J. Cogn. Neuro-

sci. 16 (4), 654–664.

Gundel, H., O’Connor, M.F., et al., 2003. Functional neuroanatomy of

grief: an fMRI study. Am. J. Psychiatry 160 (11), 1946–1953.

Habel, U., Klein, M., et al., 2005. Same or different? Neural correlates of

happy and sad mood in healthy males. NeuroImage 26 (1), 206–214.

Hamann, S., 2001. Cognitive and neural mechanisms of emotional memory.

Trends Cogn. Sci. 5 (9), 394–400.

Hamann, S.B., Ely, T.D., et al., 1999. Amygdala activity related to

enhanced memory for pleasant and aversive stimuli. Nat. Neurosci. 2

(3), 289–293.

Hariri, A.R., Bookheimer, S.Y., et al., 2000. Modulating emotional

responses: effects of a neocortical network on the limbic system.

NeuroReport 11 (1), 43–48.

Holmes, A.P., Friston, K.J., 1998. Generalizability, random effects, and

population inference. NeuroImage, 754.

Hornak, J., Bramham, J., et al., 2003. Changes in emotion after circum-

scribed surgical lesions of the orbitofrontal and cingulate cortices. Brain

126 (Pt 7), 1691–1712.

Hurlburt, R.T., Heavey, C.L., 2001. Telling what we know: describing inner

experience. Trends Cogn. Sci. 5 (9), 400–403.

Jackson, O. III, Schacter, D.L., 2004. Encoding activity in anterior medial

temporal lobe supports subsequent associative recognition. NeuroImage

21 (1), 456–462.

Ketter, T.A., Andreason, P.J., et al., 1996. Anterior paralimbic mediation of

procaine-induced emotional and psychosensory experiences. Arch. Gen.

Psychiatry 53 (1), 59–69.

Kikyo, H., Ohki, K., et al., 2002. Neural correlates for feeling-of-knowing:

an fMRI parametric analysis. Neuron 36 (1), 177–186.

Kilpatrick, L., Cahill, L., 2003. Amygdala modulation of parahippocampal

and frontal regions during emotionally influenced memory storage.

NeuroImage 20 (4), 2091–2099.

Kringelbach, M.L., O’Doherty, J., et al., 2003. Activation of the human

orbitofrontal cortex to a liquid food stimulus is correlated with its

subjective pleasantness. Cereb. Cortex 13 (10), 1064–1071.

Lane, R., 2000. Neural correlates of conscious emotional experience. In:

Lane III, R., Nadel, L. (Eds.), Cognitive Neuroscience of Emotion.

Oxford Univ. Press, New York.

A.S. Garrett, R.J. Maddock / NeuroImage 33 (2006) 263–274274

Lane, R.D., Reiman, E.M., et al., 1997. Neuroanatomical cor

relates of happiness, sadness, and disgust. Am. J. Psychiatry 154

(7), 926–933.

Lane, R.D., Reiman, E.M., et al., 1998. Neural correlates of levels of

emotional awareness. Evidence of an interaction between emotion and

attention in the anterior cingulate cortex. J. Cogn. Neurosci. 10 (4),

525–535.

Lang, P.J., Bradley, M.M., et al., 1997. International Affective

Picture System (IAPS): Technical Manual and Affective Ratings.

NIMH Center for the Study of Emotion and Attention, Gains-

ville, FL.

LeDoux, J., 2000. Listen to the brain. In: Lane, R., Nadel, L. (Eds.),

Cognitive Neuroscience. Oxford Univ. Press, New York.

Levesque, J., Eugene, F., et al., 2003. Neural circuitry underlying voluntary

suppression of sadness. Biol. Psychiatry 53 (6), 502–510.

Liberzon, I., Taylor, S.F., et al., 2000. Limbic activation and psycho-

physiologic responses to aversive visual stimuli. Interaction with

cognitive task. Neuropsychopharmacology 23 (5), 508–516.

Lieberman, M.D., Jarcho, J.M., et al., 2004. The neural correlates

of placebo effects: a disruption account. NeuroImage 22 (1),

447–455.

Maddock, R.J., Garrett, A.S., et al., 2003. Posterior cingulate cortex

activation by emotional words: fMRI evidence from a valence decision

task. Hum. Brain Mapp. 18 (1), 30–41.

Maratos, E.J., Dolan, R.J., et al., 2001. Neural activity associated with

episodic memory for emotional context. Neuropsychologia 39 (9),

910–920.

Maren, S., 2001. Neurobiology of Pavlovian fear conditioning. Annu. Rev.

Neurosci. 24, 897–931.

Maril, A., Wagner, A.D., et al., 2001. On the tip of the tongue: an event-

related fMRI study of semantic retrieval failure and cognitive conflict.

Neuron 31 (4), 653–660.

Matus-Amat, P., Higgins, E.A., et al., 2004. The role of the dorsal

hippocampus in the acquisition and retrieval of context memory

representations. J. Neurosci. 24 (10), 2431–2439.

Mayberg, H.S., Liotti, M., et al., 1999. Reciprocal limbic–cortical function

and negative mood: converging PET findings in depression and normal

sadness. Am. J. Psychiatry 156 (5), 675–682.

McGaugh, J.L., 2000. Memory-a century of consolidation. Science 287

(5451), 248–251.

McGaugh, J.L., Cahill, L., et al., 1996. Involvement of the amygdala in

memory storage: interaction with other brain systems. Proc. Natl. Acad.

Sci. U. S. A. 93 (24), 13508–13514.

Nitschke, J.B., Nelson, E.E., et al., 2004. Orbitofrontal cortex tracks

positive mood in mothers viewing pictures of their newborn infants.

NeuroImage 21 (2), 583–592.

Ochsner, K.N., Ray, R.D., et al., 2004. For better or for worse: neural

systems supporting the cognitive down- and up-regulation of negative

emotion. NeuroImage 23 (2), 483–499.

Pare, D., 2003. Role of the basolateral amygdala in memory consolidation.

Prog. Neurobiol. 70 (5), 409–420.

Phan, K.L., Wager, T., et al., 2002. Functional neuroanatomy of emotion: a

meta-analysis of emotion activation studies in PET and fMRI. Neuro-

Image 16 (2), 331–348.

Phillips, M.L., Young, A.W., et al., 1998. Neural responses to facial and

vocal expressions of fear and disgust. Proc. R. Soc. Lond., B Biol. Sci.

265 (1408), 1809–1817.

Phillips, M.L., Drevets, W.C., et al., 2003. Neurobiology of emotion

perception: I. The neural basis of normal emotion perception. Biol.

Psychiatry 54 (5), 504–514.

Pizzagalli, D., Regard, M., et al., 1999. Rapid emotional face processing in

the human right and left brain hemispheres: an ERP study. NeuroReport

10 (13), 2691–2698.

Poline, J.B., Worsley, K.J., et al., 1997. Combining spatial extent and peak

intensity to test for activations in functional imaging. NeuroImage 5 (2),

83–96.

Price, J.L., 1999. Prefrontal cortical networks related to visceral function

and mood. Ann. N. Y. Acad. Sci. 877, 383–396.

Ranganath, C., Johnson, M.K., et al., 2003. Prefrontal activity associated

with working memory and episodic long-term memory. Neuropsycho-

logia 41 (3), 378–389.

Richardson, M.P., Strange, B.A., et al., 2004. Encoding of emotional

memories depends on amygdala and hippocampus and their interac-

tions. Nat. Neurosci. 7 (3), 278–285.

Rolls, E.T., O’Doherty, J., et al., 2003. Representations of pleasant and

painful touch in the human orbitofrontal and cingulate cortices. Cereb.

Cortex 13 (3), 308–317.

Sacchetti, B., Scelfo, B., et al., 2005. The cerebellum: synaptic changes and

fear conditioning. Neuroscientist 11 (3), 217–227.

Schaefer, S.M., Jackson, D.C., Davidson, R.J., Aguirre, G.K., Kimberg,

D.Y., Thompson-Schill, S.L., 2002 Aug 15. Modulation of amygdalar

activity by the conscious regulation of negative emotion. J. Cogn.

Neurosci. 14 (6), 913–921.

Schneider, F., Habel, U., et al., 2001. Subjective ratings of pain correlate

with subcortical – limbic blood flow: an fMRI study. Neuropsychobio-

logy 43 (3), 175–185.

Schooler, J.W., 2002. Re-representing consciousness: dissociations between

experience and meta-consciousness. Trends Cogn. Sci. 6 (8), 339–344.

Sirota, A.D., Schwartz, G.E., et al., 1987. Facial muscle activity during

induced mood states: differential growth and carry-over of elated versus

depressed patterns. Psychophysiology 24 (6), 691–699.

Small, D.M., Zatorre, R.J., et al., 2001. Changes in brain activity related

to eating chocolate: from pleasure to aversion. Brain 124 (Pt 9),

1720–1733.

Small, D.M., Gregory, M.D., et al., 2003. Dissociation of neural

representation of intensity and affective valuation in human gustation.

Neuron 39 (4), 701–711.

Stark, C.E., Okado, Y., 2003. Making memories without trying: medial

temporal lobe activity associated with incidental memory formation

during recognition. J. Neurosci. 23 (17), 6748–6753.

Steele, J.D., Lawrie, S.M., 2004. Segregation of cognitive and emotional

function in the prefrontal cortex: a stereotactic meta-analysis. Neuro-

Image 21 (3), 868–875.

Strange, B.A., Otten, L.J., et al., 2002. Dissociable human perirhinal,

hippocampal, and parahippocampal roles during verbal encoding.

J. Neurosci. 22 (2), 523–528.

Taylor, S.F., Phan, K.L., et al., 2003. Subjective rating of emotionally

salient stimuli modulates neural activity. NeuroImage 18 (3), 650–659.

Visscher, K.M., Miezin, F.M., et al., 2003. Mixed blocked/event-related

designs separate transient and sustained activity in fMRI. NeuroImage

19 (4), 1694–1708.

Wager, T.D., Smith, E.E., 2003. Neuroimaging studies of working memory:

a meta-analysis. Cogn. Affect Behav. Neurosci. 3 (4), 255–274.

Wager, T.D., Phan, K.L., et al., 2003. Valence, gender, and lateralization of

functional brain anatomy in emotion: a meta-analysis of findings from

neuroimaging. NeuroImage 19 (3), 513–531.

Wagner, A.D., Pare-Blagoev, E.J., et al., 2001. Recovering meaning: left

prefrontal cortex guides controlled semantic retrieval. Neuron 31 (2),

329–338.

Wig, G.S., Miller, M.B., et al., 2004. Separable routes to human memory

formation: dissociating task and material contributions in the prefrontal

cortex. J. Cogn. Neurosci. 16 (1), 139–148.

Williams, L.M., Phillips, M.L., et al., 2001. Arousal dissociates

amygdala and hippocampal fear responses: evidence from simulta-

neous fMRI and skin conductance recording. NeuroImage 14 (5),

1070–1079.